Property underwriting data describes a location and the building on it: where it is, what hazards it faces, how it is built and protected, and how much the carrier already has nearby. A property risk score condenses some of that into a number. The value of either depends on how precisely the address is located and how closely the data describes that one property rather than its ZIP code.
The five kinds of property data
| Kind | Examples | What it answers |
|---|---|---|
| Location | Geocoded coordinates, parcel boundary, distance to the coast, a fire station or wildland vegetation | Where exactly is the property, and what is around it? |
| Hazard | Wildfire, flood, hail, tornado, wind, earthquake and storm surge, plus crime in some regions | Which perils could cause a loss here, and how likely are they? |
| Building | Year built, construction, stories, floor area, roof type and age, occupancy, replacement cost | How will this structure behave when a peril arrives? |
| Protection | Fire department and hydrant distance, sprinklers, alarms, defensible space | How quickly is a loss stopped once it starts? |
| Portfolio context | The carrier’s insured value nearby, prior claims, guidelines and moratoriums in force | Can we add this risk, and on what terms? |
Hazard and building data describe the risk. Portfolio context decides whether the carrier can take it, which is why it belongs on the same screen.
What a property risk score is
A property risk score turns hazard data, and sometimes building data, into a number or a level for one location. Scores are usually given per peril, such as a wildfire score or a hail score, because a composite score hides which peril drives the risk and each peril is priced and reinsured differently.
A useful score says what it measures (likelihood, severity or both), over what period, and against what baseline. UrbanStat’s wildfire risk score runs from 0 to 100, with a level, a long-term probability, the location’s wildfire history and real-time alerts.
How to judge a property data source
- Geocoding precision: is the address placed on the building or the parcel, or on a street segment or ZIP code centroid?
- Resolution: does the data describe this property, or an average for its ZIP code or county? The guide to ZIP codes versus property-level risk shows how much that average hides.
- Coverage and fill rate: for what share of the book does the source return a value, and what happens when it does not?
- Freshness: how often is the data updated, and are recent permits and roof replacements reflected?
- Validation: has the model been tested blind against later losses, with the results published? See wildfire model performance for one example.
- Explainability and compliance: can the underwriter see why a score is high, and does the source avoid personal data and credit information?
Residential and commercial property
For homeowners and other residential lines, the building questions are mostly about roof, age, construction and replacement cost, which decides whether the home is insured to value, and the location questions about wildfire, flood, wind and hail. Most of the answers can be filled from data without asking the applicant.
Commercial property adds what happens inside the building. Occupancy drives fire load, business interruption and liability, which is why commercial underwriters start from construction, occupancy, protection and exposure (COPE), the business type, and the statement of values (SOV) listing each location with its construction, occupancy and values. UrbanStat’s industry risk guides cover 747 business types by NAICS and SIC code, each with a COPE summary and a submission checklist.
How UrbanStat delivers property data
Typing an address, a policy number or coordinates returns more than 300 data points. The same data is available through the Risk Analysis API, where one call returns more than 100 results in under 5 seconds, in batch for whole portfolios, or in the online portal for carriers that prefer not to integrate. In the US, parcel and permit data adds construction date, building value, other structures value and building permits. Coverage by region is on data coverage.
Common questions
What is a property risk score?
A number or level that rates how exposed one location is to a peril, such as wildfire, flood or hail. Good scores are given per peril, say what period they cover, and can be traced back to the data behind them.
What data do underwriters need for a homeowners policy?
The location and its hazards (wildfire, flood, wind, hail), the building (year built, construction, roof type and age, floor area, replacement cost), protection (fire station and hydrant distance, alarms) and the carrier’s existing exposure nearby.
How do underwriters find the age of a roof?
From the application, a prior inspection, aerial imagery or building permit records. Permits are the most dependable public record of when a roof was replaced, and UrbanStat’s parcel and permit data includes building permits across all 50 states.
What is the difference between hazard data and property data?
Hazard data describes the perils at a location, whoever owns it. Property data describes the building and its use. Underwriting needs both: the same hailstorm does very different damage to a new metal roof and an old shingle one.
What is COPE in underwriting?
Construction, occupancy, protection and exposure: the four headings commercial property underwriters use to describe a risk. Each of UrbanStat’s industry risk guides includes a COPE summary for its business type.